-320-

Inzinerine Ekonomika-Engineering Economics, 2014, 25(3), 320–332

Two-Step Method Useful For Support of Technical Benchmarking Practice in the

Automotive Market

Corrado lo Storto

Department of Industrial Engineering, University of Naples Federico II

Piazzale V. Tecchio, 80 - 80125 Napoli, Italy

E-mail: corrado.lostorto@unina.it

http://dx.doi.org/10.5755/j01.ee.25.3.2659

The aim of this paper is to present a two-step method useful for support of product benchmarking practice in the

automotive industry by measuring a technical value of car model. This method assumes that a car is a bundle of

objectively and subjectively measurable attributes (i.e. functional features) provided to users. The car’s technical value is

thus a measure of the overall benefit it delivers to users, while the car’s technical efficiency measures a relative benefit the

users gain when they utilize a particular car model charged by certain ownership and usage costs. Technical efficiency is

calculated by implementing Data Envelopment Analysis (DEA). As an example, the method is implemented to conduct a

retrospective benchmarking study in the Italian domestic passenger car market in the years 1970s-1990s. Results show

that cars differed remarkably due to their technical efficiency, but only 35 car models in the sample have been classified

by DEA as 100 % efficient. Car models sold in the 1980s resulted not so competitive in terms of technical efficiency as

models sold in the 1970s and 1990s. The results also revealed that the technical value increases with the purchasing cost,

but it diminishes with the usage cost.

As additional value to literature and practice, the method suggests insights about how: a) to compare cars in a multi-

dimension features space; b) to analyze technological trends in the car industry; c) to study the car market structure and

identify the emergence of market niches still unexploited by automotive manufacturers.

Keywords: data envelopment analysis, car, benchmarking, efficiency, product value, purchasing price, cost of usage.

Introduction

Since the mid of the last century, in all developed

countries, automotive industry has played a leading role

and become a core industry for the creation of wealth and

growth of the economical systems. Automotive industry

went through an intense transformation to meet the

challenge that is coming from the market in terms of

customers’ preferences, competition, need for

manufacturing and development cost and time

compression, environmental concerns, more pressing

safety regulations, and opportunities offered by technology

advancement (Calabrese, 2009). As a consequence,

innovation in the automotive industry has acquired major

significance, with an intensive effort of OEMs to develop

new more performing car models, increase comfort and

passengers safety, reduce manufacturing costs and fuel

consumption, meet challenging environment needs, and a

differentiated consumer demand (Goldberg, 1995; Klepper,

2002). In the last decades, due to this continuous pressure

to innovate, increasing product complexity and rapid

technology progress, the amount of R&D expenses that

OEMs have to budget every year has achieved about 4.5%

of total costs in their profit and loss statement, while about

5 % of the final market price of a car model accounts for

R&D costs. Statistics also show that about 40% of total

R&D expenses are absorbed by car models which are

unable to achieve the targeted business revenue (Oliver

Wyman Automotive, 2012). For these reasons, a sound

technical or product benchmarking practice can be a

valuable means that might assist car manufacturing

companies to improve their innovative performance

identifying trajectories for improving products and make

them more competitive and appealing in the market

(Griffin, 1997; Neely, 1999; Shetty, 1993). Product

benchmarking is carried out in companies to compare the

characteristics and performance of products they sell in the

market with those of excellent competing companies with

the aim to evaluate the state-of-the-art of the embodied

technology, improve their design, manufacturing process,

and marketing strategy and, finally, achieving competitive

advantage (Lema & Price, 1995; Schumann, 1996). A

major issue in the benchmarking analysis is identification

of benchmark measurements, e.g. the standards of

excellence against which to measure and compare product

characteristics performance and carry on performance gap

analysis (Bowman & Faulkner, 1994). Within the product

development and manufacturing environment engineers

and technical managers implement different approaches

and adopt several tools that help collecting data and

process information for product benchmarking purposes,

i.e. Reverse Engineering, patent analysis and mapping,

QFD, Taguchi DOEs, FMEA, DFMA, AHP (Bergquist &

Abeysekera, 1996; Bradley & Guerrero, 2011; Hauser &

Clausing, 1988; lo Storto, 2006; Nair, 1992; Tsui, 1992;

Otto & Antonsson, 1993; Partovi, 1994; Samuelson &

Scotchmer, 2002). However, many times collecting of data

and implementing of benchmarking analysis may be a very

costly and time consuming activity.

Corrado lo Storto. Two-Step Method Useful For Support of Technical Benchmarking Practice in the Automotive Market

- 321 -

The aim of this paper is to present a two-step method

that supports benchmarking practice in the automotive

industry in the preliminary phase of the development

process using data freely available and expert judgments.

This method measures technical value of a car model

and investigates how this is associated to some economic

variables, i.e. the purchasing price and cost of usage of the

car. Measuring the value that a product delivers to users

and investigating its determinants is of paramount

importance to design and sell in the market the products

that meet the consumers’ expectation (Maleki et al., 2013).

In the first step, the method implements Data

Envelopment Analysis (DEA) to calculate relative technical

efficiency of a car model (CTE) as a weighted benefit to

cost ratio where benefits are measured by a set of

functional features performance measurements (i.e., engine

performance, quality, etc.) and costs are the car purchasing

and usage costs. A sample of car models is used to

generate a benchmark for comparison. In the second step, a

regression analysis is run to investigate the existence of

any relationships between the technical value measure of

cars in the sample and the economic variables associated

to them, using CTE measure to split sample into

meaningful groups. The technical value is measured as a

function of the benefits provided by car. Henceforth, while

in the first step benchmarking takes into account single car

models, even though each one is compared to the others or

a reduced number of them, in the second step

benchmarking is aimed at investigating general trends.

In terms of additional value to literature and practice,

the suggested method provides useful insights as to: a) how

to compare cars in a multi-dimension features space; b)

how to compare cars in terms of the objective technical

value delivered to customers; c) how to analyze

technological trends in the car industry; d) how to study

the car market structure and identify the emergence of

market niches still unexploited by automotive

manufacturers. Moreover, this method is flexible and its

implementation can be easily extended to other industries

such as aircraft, computers and printers, household

appliances, cellular phones, etc.

This paper is organized as follows. After the

introductory issues in the first section, the second section

presents the general framework of the benchmarking

method and explains how variables are measured. In this

section, steps 1 and 2 of the method are illustrated, too.

The last two sections show the results relative to the

implementation of the method in the Italian domestic car

market in the years 1970s-1990s, and present some

concluding remarks.

The Method

A General Framework

In the method, a car is conceptualised as a set of

technological features that deliver measurable

functionalities to the users. Technological features relate to

what a product is, while functional features relate to what a

product does (Saviotti & Metcalfe, 1984).

Technological features include all technological

subsystems and components that are embodied in the

tangible products. For instance, for a car these features are

engine type, suspension type, transmission system, air

conditioning equipment, etc. All subsystems and

components are working according to certain scientific

principles and design rules. These technological features

are the outcome of the choices of engineers, technological

advancement, and best engineering practices adopted in the

automotive industry at the time a car model is developed.

Functional features include the set of functions the product

delivers to the users. For instance, for a car these functions

are mobility, safety, comfort, quality, etc. Usually, one

technological feature affects more than one functional

feature. So, there is no one-to-one mapping between the

two sets of technological and functional features.

This conceptualisation of a car is consistent with what

happens in the market. Indeed, Lancaster (1966, 1971)

suggests that consumers choose and buy product

characteristics rather than products themselves that are

considered just as black boxes. Often, the users are fully

ignorant of the technological components and systems

embodied in a car and how these work integrated together.

Thus, for passenger car users it is not relevant if a four or

six cylinder engine is assembled in their car, but engine

power or speed are surely more important when they

choose a particular car model and decide to buy it. As in

the model suggested by Saviotti & Metcalfe (1984), the

method proposed in this paper takes into account

technological features (TF) and functional features (FF) of

a car, but uses the set of functional performance (FP) that

is associated to the functional features of a car, which are

more easily measurable than technological features. The

functional performance items are finally clustered into a

number of functional performance categories (FPC) that

measure the performance of groups of homogeneous

functionalities delivered by a car to the users (Figure 1,a).

Measurements for these features give a quantitative

indication of the benefits offered to car users and, at the

same time indirectly, of the nature of the technology

embodied in a car model. The technical value of a car

model (CTV) is thus assumed to be a function of the

functional performance category set (FPCi) associated to

functional features FF1, FF2, ..., FFm:

CTV = f(FPC1, FPC2, ..., FPCs) (1)

CTV is thus a measure of the overall benefit a car

delivers to users. When using a car, the consumers are also

concerned with the price they have to pay for car

availability, i.e. the ownership price of the product, and the

cost they have to bear to use the car. The decision to buy a

car is thus influenced by the product benefit/cost ratio. The

overall car technical cost (CTC) that users have to bear to

benefit by functional features FF1, FF2, …, FFm is a

function of the amount of these partial costs C1, C2, …, Cp

CTC = g(C1, C2, ..., Cp) (2)

The technical efficiency of a car can be measured as

the ratio of CTV to CTC measurements (Fig. 1,b):

CTVCTE=

CTC (3)

Inzinerine Ekonomika-Engineering Economics, 2014, 25(3), 320–332

- 322 -

b)

overall car

technical cost (CTC)

car technical

efficiency (CTE)

cost categories

CTV

functional performance

categories (FPC)

car technical

value (CTV)

CTC

FPC1

FPC2

FPC3

FPCs

. . .

C1

C2

C3

Cp

. . .

a)TF1

TF2

TF3

…

…

TFn-1

TFn

FF1

FF2

FF3

…

…

FFm-1

FFm

FP1

FP2

FP3

…

…

FPr-1

FPr

FPC1

FPC2

...

…

…

FPCs-1

FPCs

technological

features (TF)

functional

features (FF)

functional

performance set (FP)

functional performance

categories (FPC)

Figure 1. The framework

Thus, a car’s technical efficiency measures the relative

benefit the users gain when they are using a particular car

model charged by certain costs.

Table 1

Variables used for measuring benefits and costs

The Measurement of Variables

For convenience, it is assumed that both CTV and

CTC can be formulated respectively as the weighted

summation of the normalized measurements of a set of

four functional features performance, ENGINE,

MOBILITY, SAFETY, and QUALITY, and the weighted

sum of the normalized measurements of the car ownership

price, PPC, and usage cost, PUC (Table 1). Particularly, as

to the measurement of benefits, the ENGINE variable is

used as a proxy for measuring the performance of the car

engine; it depends on the engine power, engine torque,

engine capacity, and car mass. The MOBILITY variable is

a proxy which measures the car moving performance in

terms of its maximum speed, acceleration and pick up

functional features. In order to take into account several

operational conditions of a car, this variable is measured

by averaging several measurements relative to the

acceleration and pick up car performance. The SAFETY

measures the performance of the capability of a car to

assure safety for the passengers and pedestrians. The

measurement of this variable requests both subjective and

objective evaluation of the braking performance, safety

equipment, and driving quality of a car. The QUALITY

functional performance variable relates to the extent a car

provides the users (passengers and driver) with expected

quality standards as to the internal fittings, comfort,

driving easiness, etc. As to the measurement of costs, the

car purchasing price, PPC, is used as a measure of the

ownership cost, while fuel consumption, PUC, is used for

measuring the cost of product usage. As to the cost of car

usage, the operational costs of a car which are reported in

trade literature are calculated assuming that the driver will

travel a fixed distance every year (i.e., 5,000 km, 10,000

km, or 20,000 km), and include the product depreciation

too; the consequence is a strong correlation between the

purchasing price of a car and the operational cost on one

side. Moreover, the cost of car usage might be

exaggeratedly affected by the oil price. Consequently, to

avoid any bias due to correlation between variables and

market context variables, fuel consumption may be

conveniently adopted as a proxy for measuring the cost of

product usage. Major details are reported in Table A.1 in

the Appendix.

functional performance set

Benefits

ENGINE engine power, torque, capacity, car mass

MOBILITY max speed, acceleration, pick up

SAFETY braking space and quality, safety equipment

QUALITY noise, comfort, internal fittings, space, ventilation,

equipment, driving

Costs

PPC car price

PUC (average) fuel consumption

Corrado lo Storto. Two-Step Method Useful For Support of Technical Benchmarking Practice in the Automotive Market

- 323 -

Step 1: measuring the car model technical efficiency

Data Envelopment Analysis (DEA) is used to calculate

technical efficiency rate (CTE) of a car model. DEA is a

flexible non-parametric linear programming method

developed as a reformulation of the Farrell (1957)

efficiency measure to the multiple-output, multiple-input

case that evaluates relative efficiency of a number of units,

comparing the levels of inputs and outputs of one unit with

its competitors, and generating a discrete piece-wise

frontier determined by a set of efficient reference units

(Charnes et al., 1978). A unit (here, a car model) is

considered technically 100 % efficient when any other unit

uses a larger quantity of at least one of the input factors to

achieve the same output amount. Efficient cars with

“unusual” combinations cannot be directly compared to a

reference car. A car model is found to be inefficient if it is

possible to construct a “virtual” reference car as a linear

combination of other cars, such as the virtual car produces

at least the same amount of performance outputs while it

uses a lower amount of inputs than the real car under

examination. As an optimization method, DEA neither relies

on the traditional assumptions required by many other types

of analysis such as regression, nor requires any explicit

specification of underlying functional relationship that links

inputs to outputs or any weights to be assigned a priori.

The example in Figure 2 graphically illustrates how

DEA works and measures technical efficiency of units. For

simplicity, three units - Unit A, Unit B, and Unit C - are

compared, and two outputs O1 and O2 and one input I1 are

respectively produced and consumed by each unit. For

further simplification, let us assume that each unit uses the

same amount of input I1 and that the measure of such

amount is 10. The measures of the output produced by the

three units are as follows: Unit A (O1=180, O2=35), Unit B

(O1=90, O2=45), Unit C (O1=40, O2=105).

18040 90 O1

O2

A

V

B

C

45

105

35

O

Figure 2. How DEA works

DEA determines if there exists a virtual unit that

performs better than one or more of the real three units in

the example. The line segment linking Unit A to Unit C is

called the efficient frontier under the assumption of

convexity. This frontier defines the maximum

combinations of outputs that can be produced for the

assigned input. Indeed, the segment AC lies beyond both

the segment AB that can be drawn between Unit A and

Unit B and segment BC drawn between Unit B and Unit C.

As a consequence, a convex combination of Unit A and

Unit C has the capability to generate the most output for a

given set of inputs. Since Units A and C lie on the efficient

frontier they are considered 100% efficient, but as Unit B

lies under the efficient frontier, it is considered inefficient

and its efficiency (or inefficiency) can be measured as the

ratio OB/OV, where V is a virtual Unit formed through a

combination of Unit A and Unit C. Units A and C are the

reference set for Unit B. The efficiency rate of B is 70,7 %,

while it is 100% for both A and C.

DEA has revealed to be a very useful method in the

practice of benchmarking, as it provides insights as to the

potential improvement capabilities of a unit, indicates

sources of inefficiency, and also makes it possible to take

into account the existence of preferences when efficiency

rates are calculated (lo Storto, 2013; lo Storto & Ferruzzi,

2013).

DEA supplies several information that can be used in

the benchmarking analysis: a) a relative rating of products

classifying them as “efficient” or “inefficient”; b) the

reference set for each inefficient product, that is the set of

relatively efficient products to which it has been most

directly compared in calculating its efficiency rating; c) the

relative amount of specific inputs over-utilized or outputs

under-produced by inefficient products.

Several DEA models are available for measuring unit

efficiencies (Cooper et al., 2006). The proposed method

uses BCC DEA model introduced by Banker et al. (1984)

which allows taking into account scale economies due to

size difference. As car models compared in the

benchmarking study can be very different, one might

suspect that the relationship between inputs and outputs

involves variable returns to scale, i.e., that there exists a

varying relationship between increasing output and input.

The formulation of the input-oriented DEA BCC

model in the envelopment form is as follows (Cooper et

al., 2006):

, λ

0

min

subject to - 0

= 1

0,

BB

B x X

0Y y

e

(4)

where

X = (xj) Rm x n

and Y = (yj) Rs x n

are a given data set,

Rn is a column vector with all elements non-negative, e

is a row vector with all elements equal to 1, and ΘB is a

scalar.

Step 2: investigating the relationship between CTV,

PUC, and PPC

Even though BCC DEA model does not allow to have

a ranking of car models based on the calculated technical

efficiency rate, this later can be used to cluster car models

into homogeneous groups as to the efficiency score, i.e. a

group including only 100% efficient car models, and a

number of remaining groups that contain not efficient car

models. At this step of the analysis, the measurements of the

4 functional features performance categories, ENGINE,

MOBILITY, SAFETY, and QUALITY are aggregated to

get an unweighted measure of the car technical value uCTV.

The quadratic mean is calculated to aggregate measurements

of the four functional performance variables, as the

Inzinerine Ekonomika-Engineering Economics, 2014, 25(3), 320–332

- 324 -

quadratic mean is particularly sensitive either to high or low

values, thus making it possible to emphasize evident

differences. At this stage of the benchmarking analysis, the

relationships between the dependent variable uCTV, and the

independent variables PUC and PPC can be investigated

adopting the technical efficiency score class as a moderating

variable of this relationship. This analysis is complementary

to the analysis performed in step 1. As in step 1, the Cost of

Product Usage (PUC) and the Product Purchasing Cost

(PPC) can be imagined as resources that the users have to

give up to use a car model and to enjoy its functionalities

that provide them with a benefit measured by the

unweighted Car Technical Value (uCTV). Thus, both PUC

and PPC variables can be considered as factors of a

production function that produces value (uCTV) to the car

user. A convenient way to identify a formal relationship

linking together these variables is to use the Cobb-Douglas

formulation (Bridge, 1971; Cobb & Douglas, 1928;

Richmond, 1974):

b cauCTV PUC PPC (5)

Here a is a constant, which depends on the units in

which inputs and outputs are measured, while b and c are

constants that take into account the relative importance of

PUC and PPC in delivering technical value uCTV to car

users. No particular assumptions or constraints relative to

values assumed by these constants are imposed in the

estimation of these parameters.

An Example: the Italian Domestic Car Market

from the 1970s to 1990s

The Italian passenger car domestic market from the

early 1970s to the 1990s was considered to implement the

benchmarking method. All the car models selected for the

analysis were ordinary passenger cars that have been

equipped with conventional spark ignition petrol engines

or turbocharged spark ignition engines. Data relative to

cars have been collected from trade literature having as a

reference three temporal market windows, the 1970-72s

market (37 car models), the 1980-82s market (82 car

models), and the 1990-93s market (97 car models). Each

sub-sample was selected with the aim of having a good

mix of all passenger cars sold in Italy in that period. The

data were collected from trade literature (GenteMotori,

1980 to 1993; Quattroruote, 1970 to 1993). The use of

published data found in the automotive press has the

advantage of standardization, completeness, and

impartiality of measurement. However, the need to assess

and compare subjectively features required the researcher

to consult an expert of the automotive field. In this case, a

five points Likert-type scale was used to measure

functional features’ performance that could be measured

only by means of subjective expert judgements (Table A.1

in Appendix). Moreover, to get comparable data, the

purchasing prices for all cars were measured with

reference to the year 1993 using the consumer price index

CPI(1993). Next, the results relative to step 1 and step 2

are illustrated.

Table 2

The outcome of DEA: means of variables across groups*

variable G1 CTE=100%

# 35

G2 82,91%<CTE<100%

# 90

G3 CTE<=82,91%

# 91

engine capacity (cc) 2,070,31

[1,359]

1,514,91

[642,71]

1,774,59

[618,36]

car price (€, CPI 1993) 26,055

[34,161]

14,620

[15,390]

15,568

[12,738]

# cars in the 1970s 5 (14%) 19 (51%) 13 (35%)

# cars in the 1980s 3 (4%) 22 (27%) 57 (69%)

# cars in the 1990s 27 (28%) 49 (51%) 21 (22%)

efficiency score 100% 89,99%

[4,67%]

73,17%

[7,60%]

PUC 0,470

[0,156]

0,412

[0,077]

0,476

[0,085]

uCTV 0,626

[0,104]

0,533

[0,088]

0,535

[0,057]

ENGINE 0,463

[0,154]

0,382

[0,112]

0,383

[0,074]

MOBILITY 0,479

[0,142]

0,393

[0,088]

0,408

[0,064]

SAFETY 0,704

[0,129]

0,595

[0,115]

0,565

[0,085]

QUALITY 0,772

[0,123]

0,689

[0,114]

0,708

[0,075]

*In square brackets standard deviations and in round brackets percentages are respectively indicated

Corrado lo Storto. Two-Step Method Useful For Support of Technical Benchmarking Practice in the Automotive Market

- 325 -

Step 1

Table 2 shows the outcome of DEA. The sample was

split into three groups, depending on technical efficiency

score. Particularly, group G1 contains 100 % efficient car

models. The second group G2 contains car models whose

technical efficiency rate is between 82,91 % and 100 %,

while group G3 includes car models having technical

efficiency lower than 82,91 %. The 82,91 % threshold is the

median of the smaller sample made of 181 non efficient

cars. The average sample technical efficiency is 81,53 %,

while the minimum score is 45,21 %. Thus, there is a great

variance as to efficiency of passenger cars in sample.

Table 3

The outcome of DEA: the 100 % efficient car models

car model year engine

capacity

engine

power

max

speed price

occurrence in

reference sets

type of

market

Fiat 127 1970s 903 47 143,85 5,153 61

Mazda RX2 Coupè 1970s 1,146 130 183,54 12,756 0 niche

Innocenti Mini Cooper 1300 1970s 1,275 71 160,71 7,398 12

Fiat 128 Rallye 1970s 1,290 67 152,93 6,939 5

Simca 1000 Rallye 1 1970s 1,294 60 153,03 5,505 40

Renault Alpine Turbo 1980s 1,397 110 184,5 11,386 7

Talbot Sunbeam Lotus 1980s 2,172 155 197,1 15,357 0 niche

Jaguar XJ 5.3 1980s 5,345 287 227,8 47,249 0 niche

Fiat 500 new (700) 1990s 704 30 119,76 5,060 32

Subaru M80 5P 1990s 758 42 142,75 6,027 94

Fiat 500 new (900) 1990s 903 41 136,03 5,871 11

Daihatsu Charade Gti Turbo 1990s 993 101.35 194,9 11,131 145

Autobianchi Y10 Avenue 1990s 1,108 50.5 147,88 8,547 50

Peugeot 306XT 1.4 1990s 1,360 75 165,6 11,349 7

Fiat Uno 1.4 IE sx 5P 1990s 1,372 71 166,7 9,270 9

Opel Corsa Swing 1.4 ie 3P 1990s 1,388 60 155,6 7,953 10

Renault Clio automatic 1990s 1,390 76.5 163,83 11,104 17

Opel Astra 1.6i GLS 5P 1990s 1,598 100.5 190 11,110 13

Peugeot 306XT 1990s 1,761 102.8 184,7 11,938 4

Audi 100 2.0 16V Avant 1990s 1,984 140 204,21 26,722 1

Audi 100 2.0 Ecat 1990s 1,984 115.5 188,08 23,250 6

Volvo 850 GLT L 1990s 1,984 143 202,3 20,710 2

SAAB 900 Saero 1990s 1,985 112.9 208,2 21,095 0 niche

Ford Mondeo 2000 Ghia 1990s 1,988 134 206,99 15,205 16

Ford Superescort RS luxury 1990s 1,993 223.35 224,3 28,121 0 niche

Rover 220 Turbo 1990s 1,994 200 237,8 19,548 17

Honda Accord 2.0i 16V Coupè 1990s 1,997 133 202,15 20,090 3

Nissan Sunny Gti - R 1990s 1,998 220 22,37 25,554 1

Opel Astra Gsi 16V 1990s 1,998 150 218,79 15,748 11

Maserati Ghibli 1990s 1,998 306 262,6 41,774 8

Mercedes 280E-24V 1990s 2,799 197 230 38,815 0 niche

Alfa Romeo 164 V6 24V super 1990s 2,959 210 240 33,505 1

Ferrari 512 TR 1990s 4,943 428 314 143,916 12

Lamborghini Diablo VT 1990s 5,707 492 325 157,668 0 niche

Jaguar XJR-s 1990s 5,993 333,5 253,45 69,092 1

Table 2 shows some relevant findings from DEA:

car models that are 100% technical efficient are

more expensive having a higher purchasing price (or, the

normalized PPC), even though the standard deviation of

this variable is great;

unexpectedly, most cars (69 %) in group G3 that

includes less efficient models were sold in the 1980s, while

only 4 % of passenger cars in these years is 100 % efficient;

the amount of car models sold in the market in the

1970s and belonging to group G2 is the same of cars sold

Inzinerine Ekonomika-Engineering Economics, 2014, 25(3), 320–332

- 326 -

in the 1990s, while the amount of 1980s cars in this group

remains smaller. These figures clearly make evident that

passenger cars sold in the Italian market in the 1980s were

not as competitive as cars in the 1970s and 1990s. That is

not surprising, as between the end of the 1970s and the mid

of the 1980s there was a profound restructuring of the

manufacturing and product development processes in

search of a higher production efficiency to decrease costs

and achieve better product quality. Indeed, there was a

great effort to survive competition coming from the Far

East car manufacturers, primarily from Japan. This effort

was successful as the automotive industry was able to

improve performance of both manufacturing and product

development processes;

on the average, car models in group G1 have

higher functional feature performance measures and an

overall uCTV, but – in the same time – are more

expensive, even though with a great price variance.

Table 3 illustrates some details relative to 35 car

models identified by DEA as 100 % efficient. As the

previous table showed, this group of cars is rather

variegated as it contains models that belong to several

market segments classified, for instance, as A (i.e., Fiat

127), B (Simca 1000 Rallye), or even sport cars

(Lamborghini Diablo VT). That should not be surprising,

as DEA identifies efficient units on the base of the ratio of

weighted outputs to weighted inputs. The last column but

one presents information that is useful to assess the

competitiveness of cars, i.e. the number of times each

model compares in the reference set of an inefficient car.

Seven passenger cars – Mazda RX2 Coupè, Talbot

Sunbeam Lotus, Jaguar XJ5.3, Saab 900 Saero, Ford

SuperEscort RS luxury, Mercedes 280E-24V, and

Lamborghini Diablo VT – have only themselves as a

reference car, not being in any reference set. This

information can be used to identify market niches of the

product offering. “A niche market is a relatively small

segment of a market that the major competitors or

producers may overlook, ignore, or have difficulty serving.

The niche may be a narrowly defined geographical area, it

may relate to the unique needs of a small and specific

group of customers, or it may be some narrow, highly

specialized aspect of a very broad group of customers”

(Gross et al., 1993, p. 360). Effective niche strategies may

be sometimes very profitable, because a niche market may

actually be very large. Emphasis on niche marketing

provides a very clear focus for the development of business

strategies and action plans. As a final comment about

figures in the “occurrence in reference sets” column, two

car models merit particular attention, Daihatsu Charade Gti

Turbo and Subaru M80 5P, the first one in the reference

sets of 145 cars and the second in those of 94 cars. So,

even though both cars are efficient, they occupy a market

position that clearly is not defendable. Unexpectedly, the

Ferrari 512 TR that was sold in the market in the 1990s

appears in the reference sets of 12 cars, including some

cars that do not belong to the same market segments (e.g.,

BMW 318i and BMW 730i). Of course, customers who

buy a Ferrari car do not expect to have higher technical

value as the only benefit for their expensive purchase!

The analysis of the reference sets of inefficient car

models provides insights about the nature of competition in

the market. Table 4 reports the reference sets for some

inefficient car models extracted from sample. As to the

first car in table, Volkswagen Golf 2,8 vr6, three cars of its

reference set are clearly in the same market segment (Ford

Mondeo 2000 Ghia, Rover 220 Turbo, and Alfa Romeo

164). Even, this car has as its reference a Ferrari. The

second car, Citroen Gs Club, has in its reference set two

car models sold in the market twenty years later (both Fiat

500) and one car that in the 1970s was in a higher market

segment (Fiat 128 Rallye). Two cars, Fiat Ritmo 75s and

Fiat Argenta 2000, have the same reference set made of

cars positioned in a lower market segment (A). But, the

comparison of Fiat Argenta with cars of the reference set is

much more unfavorable (as emphasized by the efficiency

score). Indeed, in the automotive market positioning Fiat

Argenta is much more distant from segment A than Fiat

Ritmo. Finally, Jaguar XJS 4.2, which is the lower

performing car in sample in terms of technical efficiency,

is compared with cars that position between the A and B

market segments, even though the reference cars appeared

in the market ten years later.

Table 4

The reference sets of some inefficient car models

car model reference set

VOLKSWAGEN Golf

2,8 vr6

Ford Mondeo 2000 Ghia, Rover 220

Turbo, Alfa Romeo 164 V6 24V super, Ferrari 512 TR

CITROEN Gs Club Fiat 127, Fiat 128 Rallye, Fiat 500 new

(700), Fiat 5oo new (900)

FIAT Ritmo 75S Subaru M80 5P, Daihatsu Charade Gti

Turbo, Autobianchi Y10 Avenue

FIAT Argenta 2000 Subaru M80 5P, Daihatsu Charade Gti Turbo, Autobianchi Y10 Avenue

JAGUAR XJS 4.2 Daihatsu Charade Gti Turbo, Renault Clio

automatic

Table 5 shows some information that further makes

evident the strength of DEA in the practice of product

benchmarking. In particular, this table illustrates how DEA

can be used to identify some improvement trajectories for

inefficient car models. The efficiency rating provided by

DEA suggests the degree of inefficiency of a car model

compared with a virtual car on the frontier defined by its

reference set. However, it does not provide any ranking of

cars. Thus, for instance, the car model Fiat Argenta 2000 is

about 63,66 % efficient compared with its reference set

cars, while Citroen Gs Club is about 82,91 % efficient if

compared with cars on its reference frontier segment (Fiat

127, Fiat 128 Rallye, and both Fiat 500 new models).

Generally, this means that Fiat Argenta should reduce the

cost of usage and purchasing price by approximately 36,34

% = 100 % - 63,66 % without decreasing the performance of

any functional features delivered to users in order to increase

its overall efficiency score. In theory, technical efficiency of

inefficient cars might be improved either by increasing the

functional performance outputs or by decreasing inputs used

(e.g., cost of usage and purchasing price).

Table 5 summarizes the DEA outcome regarding

specific inputs that inefficient cars over-utilize or outputs

that they under-produce. The extent to which inputs can be

reduced is indicated as a negative percentage by figures in

columns “PPC” and “PUC”, while extra output generated

Corrado lo Storto. Two-Step Method Useful For Support of Technical Benchmarking Practice in the Automotive Market

- 327 -

by the inefficient car moving toward the efficient frontier

as positive percentage in the remaining columns (Engine,

Mobility, Safety, and Quality) that indicate the extent to

which output benefits should be increased to move the car

to the efficient frontier. For instance, the car model Fiat

Ritmo 75s can become efficient by decreasing its

purchasing price by about 29 %. As a general rule for

decision-making, if on the average the excess of a certain

input is extremely high, that input is not critical because

there might be large room for improvement. Vice versa, if the input excess is very low, that input variable might be

seriously critical when redesigning that car model because

of a limited space of action.

Table 5

Potential improvement of functional features/potential reduction of costs for some car models

car model years CTE (%) PPC PUC ENGINE MOBILITY SAFETY QUALITY

VOLKSWAGEN Golf 2.8 vr6 1990s 89,60 -10,4 % -10,4 % 17,0 % 0 % 0 % 3,8 %

CITROEN Gs Club 1970s 82,91 -17,1 % -17,1 % 1,0 % 0 % 0 % 9,1 %

FIAT Ritmo 75S 1980s 74,29 -29,4 % -25,7 % 1,4 % 0 % 21,91 % 0 %

FIAT Argenta 2000 1980s 63,66 -42,4 % -36,4 % 0 % 0 % 31,23 % 0,77 %

JAGUAR XJS 4.2 1980s 45,21 -67,3 % -54,8 % 2,2 % 0 % 10,6 % 0 %

Table 6

The outcome of the nonlinear regression analysis

G1 (CTE = 100%) G2 (82,91% < CTE < 100%) G3 (CTE≤ 82,91%)

parameter estimate t-value p-level estimate t-value p-level estimate t-value p-level

a 0,831 32,759 0,000 0,810 27,611 0,000 0,756 32,994 0,000

b -0,370 -4,299 0,000 -0,268 -4,466 0,000 -0,224 -4,296 0,000

c 0,269 9,343 0,000 0,252 15,855 0,000 0,211 12,628 0,000

loss function final value=0,0642 loss function final value=0,1409 loss function final value=0,0914

% variance explained=83,7 % % variance explained=79,6 % % variance explained=68,5 %

R=0,915 R=0,892 R=0,828

Step 2

In step 2, benchmarking study is conducted at a more

aggregate level, in order to identify some general trends

which can guide marketing professionals, engineers and

designers in their search for a better and more successful

product.

Table 6 shows the outcome of the regression analysis

between the unweighted car technical value (uCTV) as a

dependent variable and the car cost of usage (PUC) and

purchasing price (PPC) as independent variables.

Particularly, using the equation presented in (5) and the

Livenberg-Marquardt least square estimation method, a

nonlinear regression was performed for each group. Table

6 provides information relative to parameter estimates,

statistical significance, and predictive reliability. All

parameters are significant at least 1 %, and variance

absorbed is between 68,5 % and 83,5 %. In all cases,

estimate of parameter b is negative, while estimate of c is

positive. Thus, uCTV increases when the purchasing price

(PPC) increases, and diminishes when the cost of usage

(PUC) of a car increases. The uCTV sensitiveness to PUC

increase is higher in group G1. A graphical visualization of

these relationships may better support the analysis.

Figures 2a, 2b, 2c illustrate how uCTV changes as a

function of PUC, for fixed PPC values. These plots disclose

how the investigated relationship may be affected by the

technical efficiency score. In particular, for a low purchasing

price (Figure 2,a), when PUC is greater than 0,6, inefficient

cars of group G3 seem to behave better in terms of technical

value delivered to users. But, when PUC is far below this

threshold, efficient cars in group G1 provide higher benefit.

Worth to note that cars in group G3 behave better than cars

in group G2, even being less efficient.

With a small purchasing price increase (Figure 2,b),

the PUC threshold that determines a change in the way

cars belonging to different groups behave moves ahead,

about PUC=0,8. For this purchasing price, cars in group

G2 are better than cars in group G3.

Moving to the last graph, the better behavior of cars in

group G2 compared to cars in group G3 becomes more

evident.

0.35

0.40

0.45

0.50

0.55

0.60

0.2 0.4 0.6 0.8 1.0

uC

TV

PUC

uCTV(G1) uCTV(G2) uCTV(G3)

Figure 2, a. Plot of uCTV vs PUC, PPC=0,05

Inzinerine Ekonomika-Engineering Economics, 2014, 25(3), 320–332

- 328 -

0.40

0.45

0.50

0.55

0.60

0.65

0.70

0.75

0.2 0.4 0.6 0.8 1.0

uC

TV

PUC

uCTV(G1) uCTV(G2) uCTV(G3)

Figure 2, b. Plot of uCTV vs PUC, PPC=0,10

0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.2 0.4 0.6 0.8 1.0

uC

TV

PUC

uCTV(G1) uCTV(G2) uCTV(G3)

Figure 2, c. Plot of uCTV vs PUC, PPC=0.20

Conclusion

This paper has proposed a two-step method useful for

implementation of the benchmarking practice to compare

products in the automotive market. It is assumed that a car

is a bundle of objectively and subjectively measurable

attributes or functional features delivered to the users.

In the first step, the method adopts Data Envelopment

Analysis to calculate the relative technical efficiency of a

car model (CTE) as a weighted benefits to costs ratio

where benefits are associated to a set of functional features

performance measures (Engine, Mobility, Safety, and

Quality) and costs are measured by the car purchasing and

usage costs. A sample of car models is used to generate a

benchmark for comparison. In the second step, a nonlinear

regression analysis is run to investigate the existence of

relationships between the measure of technical value of car

models and the associated economic variables, using the

CTE measure to split the sample into meaningful groups.

The technical value is measured as a function of the

benefits provided by a car. Henceforth, while in the first step

product benchmarking has a micro-analytic perspective,

focusing on single car models and taking into account the

specific measurements of their features, even though each

car is compared to the others or to a reduced number of

them, in the second step benchmarking has a macro-

analytic perspective, aimed at investigating general trends

in the market.

As an illustrative case, the method has been

implemented to benchmark a sample of 216 cars that were

sold in the Italian domestic market between the 1970s and

the early 1990s. The results show that passenger cars in the

sample differ remarkably as to their technical efficiency,

but only 35 car models have been classified by DEA as

100 % efficient. This group of efficient cars includes

models that belong to several market segments, i.e. city

cars or sport cars. Generally, 100 % efficient car models

tend to be more expensive than not efficient cars having a

higher purchasing price. In the second step of the method

the findings revealed that car’s technical value CTV

increases when the car’s purchasing price (PPC) increases,

but it diminishes when the cost of usage of a car (PUC)

goes up. Passenger car models sold in the market in the

1980s resulted not so competitive in terms of technical

efficiency as models sold in the 1970s and 1990s. The

method has also provided useful insights as to the nature of

competition in the Italian car market from 1970s to 1990s.

In particular, seven market niches have been identified.

Car models that are in these niches such as the Mazda RX2

Coupè, Talbot Sunbeam Lotus, Jaguar XJ 5.3, Saab 900

Saero, Ford SuperEscort RS luxury, Mercedes 280E-24V,

and Lamborghini Diablo VT had some specific

combinations of functional features that made them unique

car models in the market. Unexpectedly, the Ferrari 512 TR

sold in the market in the 1990s was not a market niche car,

as it was a benchmark for 12 different car models, even not

belonging to the same market segment, such as the BMW

318i and BMW 730i models. Furthermore, some car models

sold in the 1970s remained still competitive in the 1980s and

1990s, as the Citroen Gs Club, while some others were

competitive in different market segments, e.g. the Fiat Ritmo

75s. The method also suggested how to improve specific

performance categories for each inefficient car model to

make them more competitive in the market.

Even though the method has been applied to conduct a

retrospective analysis of the Italian car market, the

utilization of a set of objective and subjective metrics for

measuring performance of functional features delivered to

the users and cost parameters rather than technology features

embodied in a car makes it simple, flexible and easily

implementable to study the present country-specific car

markets worldwide. The method may easily incorporate

further functional features measurements. That is the case of

parameters that measure the extent to which a car is

environment respectful. Until the 1990s environmental

concerns were not challenging in the Italian car market, and

for this reason - given the specificity of the sample that

collects cars sold between 1970s and 1990s – the

implementation of the method has not taken into account

any environmental concerns, i.e. tailpipe emissions, carbon

dioxide production, and recycling. When a more refined

analysis is needed, all or some functional measurements

may be fragmented into their components using these later

as the outputs in DEA implementation. Finally, the

method’s flexibility allows introducing a weighting system

that accounts for the preferences that consumers have

either for certain functional categories or some functional

features by adding further constraints in DEA model

formulation.

Furthermore, application of the method can be

extended to other industries such as aircraft, computers and

printers, cellular phones, household appliances, etc.

Corrado lo Storto. Two-Step Method Useful For Support of Technical Benchmarking Practice in the Automotive Market

- 329 -

The adoption of such a method as a technical

benchmarking and product analysis tool could help

managers to make sound decisions and plans. Indeed, the

positioning of a product based on a sound benchmarking of

functional features is useful to explore the market

competitive inter-relationships among different products in

the same segment or belonging to different segments, and

to identify temporal changes in a manner that is similar in

appearance to a perceptual map. Generating measurements

for the technical value and the technical efficiency of a

product linked to its capability to provide the users with

benefits associated to a set of functional features suffering

some ownership and usage costs helps to get information

and alleviate ambiguity related to a number of issues, i.e.

an in-depth comprehension of the nature of competition

relative to certain types of product features, the relative

assessment of the whole set of product performance, how

to increase product performance by improving specific

functional and technological features, the correct product

pricing and advertising strategy, the identification of gaps

or niches within some segments of the market, etc. In

general, the comparative assessment of products provides

managers with important insights as to how products can

be improved or new product can be developed to fit more

closely with the opportunities offered by technology and

market needs, giving the company a competitive edge.

Indeed, measuring the technical efficiencies of products

and mapping their functional features may contribute to

gain insights related to current and prospective product

offering, helping to find business opportunities for

improving the existing products or launch new product in

the market. As the state of technology does not remain

static either in the short or the long run, and new

technology devices can be mounted in a product to have

better or new functionalities at disposal of the users, the

method can also be usefully adopted to implement

dynamic benchmarking studies, as it was illustrated by the

analysis of the Italian car market presented in this paper.

For instance, in the short run one way that products

compete in the market is by leapfrogging each other in

terms of performance - whether measured in speed, safety,

quality, comfort, reliability, etc. When the measurements

of the technical efficiencies of a sample of products either

in the same or in different segments are averaged and used

as a single efficiency score, the product’s technical

efficiency can be utilized to have a picture of the

technology state in that product market, or to trace the

evolution of the technical value of the product in its market

segment over time, and analyze the relationships between

performance, technology, and costs.

Of course, the benchmarking studies which adopt the

proposed method that is fundamentally based on the

analysis of product features support rather than substitute

for the perceptual data that can be provided by customers

when market demand should be analyzed (see, for

instance, Djokic et al., 2013). Methods that take into

account perceptual data remain critical to understand the

determinants of the consumers purchasing behavior and the

relation between this and their perception of product value

(Kazakeviciute & Banyte, 2012).

Appendix Table A.

Measurements of variables

1

2

i ii ENG ENG

i i

ENG

POW TORENGINE

CAP MASS

max engine power of car ii

ENGPOW , max engine torque of car ii

ENGTOR , engine capacity of car ii

ENGCAP ,

mass of car iiMASS All measurements are objective and available in trade technical literature. ENGINEi was further normalized in the range [0, 1] by dividing its

measurement by the maximum ENGINE value in sample.

1

1 1 3

6 41 2 3 4 5 6 1 2 3 4

i i i i i i i i i i i iMAX

MOBILITY AC AC AC AC AC AC U U U U V

1 acceleration of car i in the space [0 - 1 km]iAC ,

2 acceleration of car i after 400 miAC ,

3 acceleration of car i to increase speed from 0 to 60 kmhiAC ,

4 acceleration of car i to increase speed from 0 to 80 kmhiAC ,

5 acceleration of car i to increase speed from 0 to 100 kmhiAC ,

6 acceleration of car i to increase speed from 0 to 120 kmhiAC ,

1 pick up of car i to increase speed from 40 kmhiU ,

2 pick up of car i to increase speed from 70 to 80 kmhiU ,

3 pick up of car i to increase speed from 70 to 100 kmhiU ,

4 pick up of car i to increase speed from 70 to 120 kmhiU ,

4 pick up of car i to increase speed from 70 to 120 kmhiU , max speed of car i

MAXV

All measurements are objective and available in trade technical literature. MOBILITYi was further normalized in the range [0, 1] by dividing its

measurement by the maximum MOBILITY value in sample.

1

1 1 2

2 41 2 3 4

i i i i i i iSAFETY BRAS BQ S S S S

1

31 2 3

i

MAXi

i i i

MIN

MASS

MASSBRAS

BS BS BS

BS

subjective measure of the braking quality of car iiBQ , braking space of car iiBRAS , mass of car iiMASS ,

maximum car mass in sampleMAXMASS , 1

braking space of car i at speed of 60 kmhiBS ,

Inzinerine Ekonomika-Engineering Economics, 2014, 25(3), 320–332

- 330 -

2 braking space of car i at speed of 80 kmhiBS ,

3 braking space of car i at speed of 100 kmhiBS ,

3 braking space of car i at speed of 100 kmhiBS ,

1

31 2 3

minimum in sample MIN i i iBS BS BS BS ,

1 subjective measure of the steering quality of car iiS ,

2 subjective measure of the visibility quality of car iiS ,

3 subjective measure of the road holding quality of car iiS ,

4 subjective measure of the safety equipment quality of car iiS

All subjective measurements were provided by expert judgment by means of a 5 level Likert type scale in the range [0, 1]. Objective measurements were

available in trade technical literature. SAFETYi was further normalized in the range [0, 1] by dividing its measurement by the maximum SAFETY value

in sample.

1

1 1 3

4 51 2 3 4 1 2 3 4 5

i i i i i i i i i i iQUALITY NO NO NO NO IQ IQ IQ IQ IQ CO

1 internal noise level of car i at speed of 60 kmhiNO ,

2 internal noise level of car i at speed of 80 kmhiNO ,

3 internal noise level of car i at speed of 100 kmhiNO ,

4 internal noise level of car i at speed of 120 kmhiNO ,

1 subjective measure of the car i internal fittings qualityiIQ ,

2 subjective measure of the car i internal ventilation and climate qualityiIQ ,

3 subjective measure of the car i internal equipment qualityiIQ ,

4 subjective measure of the car i internal space qualityiIQ ,

5 subjective measure of the car i driving seat qualityiIQ , subjective measure of the car i travel comfort qualityiCO

All subjective measurements were provided by expert judgment by means of a 5 level Likert type scale in the range [0, 1]. Objective measurements were available in trade technical literature. QUALITYi was further normalized in the range [0, 1] by dividing its measurement by the maximum QUALITY

value in sample.

1

2 2 2 2 2

4

i i i ii ENGINE MOBILITY SAFETY QUALITY

uCTV

1

31 2 3

i i i iPUC FU FU FU

1 fuel consumption of car i in city drivingiFU ,

2 fuel consumption of car i at speed of 90 kmhiFU ,

3 fuel consumption of car i at speed of 120 kmhiFU

1993

1993

100

CP(I ) i i

tPPC PPC

1993CP(I ) the 1993 consumer price index , 1993

purchasing price of car i at year 1993iPPC ,

purchasing price of car i sold at year ti

tPPC

References

Banker, R. D, Charnes, A., & Cooper, W. W. (1984). Some models for estimating technical and scale inefficiencies in data

envelopment analysis. Management Science, 30 (9), 1078–92. http://dx.doi.org/10.1287/mnsc.30.9.1078

Bergquist, K., & Abeysekera, J. (1996). Quality function deployment (QFD) — A means for developing usable products.

International Journal of Industrial Ergonomics, 18 (4), 269–275. http://dx.doi.org/10.1016/0169-8141(95)00051-8

Bowman, C., & Faulkner, D. (1994). Measuring product advantage using competitive benchmarking and customer

perceptions. Long Range Planning, 27 (1), 119–132. http://dx.doi.org/10.1016/0024-6301(94)90012-4

Bradley, J. R., & Guerrero, H. H. (2011), An alternative FMEA method for simple and accurate ranking of failure modes.

Decision Sciences, 42 (3), August, 743–771. http://dx.doi.org/10.1111/j.1540-5915.2011.00329.x

Bridge, J. L. (1971). Applied Econometrics. Amsterdam: North-Holland Publishing Company.

Calabrese G. (2009). Editorial to the Special Issue on “Innovation Design and Sustainable Development in the Automobile

Industry”. International Journal of Automotive Technology and Management, 9(2), 111–122.

Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision-making units. European Journal

Operations Research, 2 (6), 429–444. http://dx.doi.org/10.1016/0377-2217(78)90138-8

Cobb, C. W., & Douglas, P. H. (1928). A theory of production. American Economic Review, 18 (1), 139–165.

Cooper, W. W., Seiford, L. M., & Tone, K. (2006). Introduction to Data Envelopment Analysis and its uses. New York:

Springer.

Djokic, N., Salai, S., Kovac-Znidersic, R., Djokic, I. & Gordana Tomic, G. (2013). The Use of Conjoint and Cluster

Analysis For Preference-Based Market Segmentation. Inzinerine Ekonomika-Engineering Economics, 24(4), 343–

355. http://dx.doi.org/10.5755/j01.ee.24.4.3118

Farrell, M. J (1957). The measurement of productive efficiency. Journal of the Royal Statistical Society, 120(3), 253–281.

http://dx.doi.org/10.2307/2343100

GenteMotori, Years: 1980, 1981, 1982, 1991, 1992, 1993. Milano: Rusconi Editore.

Corrado lo Storto. Two-Step Method Useful For Support of Technical Benchmarking Practice in the Automotive Market

- 331 -

Goldberg, P. K. (1995). Product differentiation and oligopoly in international markets: The case of the U.S. automobile

industry. Econometrica, 63(4), 891–951. http://dx.doi.org/10.2307/2171803

Griffin A. (1997). PDMA research on new product development practices: Updating trends and benchmarking best

practices. Journal of Product Innovation Management, 14 (6), 429–458. http://dx.doi.org/10.1016/S0737-

6782(97)00061-1

Gross, A. C., Banting, P. M., Meredith, L. N., & Ford, I. D. (1993). Business Marketing. Houghton Mifflin Company.

Hauser, J. R., & Clausing, D. (1988). The House of Quality. Harvard Business Review, May-June, 63-73.

Kazakeviciute, A. & Banyte, J. (2012). The Relationship of Consumers Perceived Hedonic Value and Behavior. Inzinerine

Ekonomika-Engineering Economics, 23(5), 532–540. http://dx.doi.org/10.5755/j01.ee.23.5.1975

Klepper, S. (2002). The capabilities of new firms and the evolution of the US automobile industry. Industrial and

Corporate Change, 11 (4), 645–666. http://dx.doi.org/10.1093/icc/11.4.645

Lancaster, K. J. (1966). A new approach to consumer theory. Journal of Political Economy, 74 (2), 132–157.

http://dx.doi.org/10.1086/259131

Lancaster, K. J. (1971). Consumer demand: A new approach. New York: Columbia University Press.

Lema, N. M., & Price, A. D. F. (1995). Benchmarking: performance improvement toward competitive advantage. Journal

of Management in Engineering, 11 (1), 28–37. http://dx.doi.org/10.1061/(ASCE)0742-597X(1995)11:1(28)

lo Storto, C. (2006). A method based on patent analysis for the investigation of technological innovation strategies: the

European medical prostheses industry. Technovation, 26(8), 932–942. http://dx.doi.org/10.1016/j.technovation.

2005.10.005

lo Storto, C. (2013). A DEA based model for ranking air freighters operational efficiency. Applied Mechanics and

Materials, 390, 155–160. http://dx.doi.org/10.4028/www.scientific.net/AMM.390.155

lo Storto, C., & Ferruzzi, G. (2013). Benchmarking economical efficiency of renewable energy power plants: A Data

Envelopment Analysis Approach. Advanced Materials Research, 772, 699–704. http://dx.doi.org/10.4028/

www.scientific.net/AMR.772.699

Maleki, M., Shevtshenko, E., & Cruz-Machado, V. (2013). Comparative Analysis of Customer Value Dimensions.

Inzinerine Ekonomika-Engineering Economics, 24(5), 488–495. http://dx.doi.org/10.5755/j01.ee.24.5.3241

Nair, V. N. (1992). Taguchi's Parameter Design: A panel discussion. Technometrics, 34(2), 127–161.

http://dx.doi.org/10.1080/00401706.1992.10484904

Neely, A. (1999). The performance measurement revolution: why now and what next. International Journal of Operations

and Production Management, 19 (2), 205–228. http://dx.doi.org/10.1108/01443579910247437

Oliver Wyman Automotive, (2012). 2015 Car innovation: A comprehensive study on innovation in the automotive

industry, http://www.e-motility.com/Oliver_Wyman_Car_Innovation.pdf (retrieved on line Aug 10, 2012).

Otto, K. N., & Antonsson, E. K. (1993). Extensions to the Taguchi method of product design. Journal of Mechanical

Design, 115 (1), 5–13. http://dx.doi.org/10.1115/1.2919325

Partovi, F. Y. (1994). Determining what to benchmark: an analytic hierarchy approach. International Journal of

Operations & Production Management, 14 (6), 25–39. http://dx.doi.org/10.1108/01443579410062068

Quattroruote, Years: 1970, 1971, 1972, 1980, 1981, 1982, 1991, 1992, 1993. Milano: Editoriale Domus.

Richmond, J. (1974). Estimating the efficiency of production. International Economic Review, 15 (2), 515–521.

http://dx.doi.org/10.2307/2525875

Samuelson, P., & Scotchmer, S. (2002). The law and economics of reverse engineering. The Yale Law Journal, 111 (7),

1575–1663. http://dx.doi.org/10.2307/797533

Saviotti, P. P., & Metcalfe, J. S. (1984). A theoretical approach to the construction of technological output indicators.

Research Policy, 13 (3), 141–151. http://dx.doi.org/10.1016/0048-7333(84)90022-2

Schumann, C. A. W. (1996). Discussion of “Benchmarking: performance improvement toward competitive advantage”.

Journal of Management in Engineering, 12 (1), 59–61. http://dx.doi.org/10.1061/(ASCE)0742-597X(1996)12:1(59)

Shetty, Y. K. (1993). Aiming high! Competitive benchmarking for superior performance. Long Range Planning, 26 (1),

39–44. http://dx.doi.org/10.1016/0024-6301(93)90231-4

Tsui, K. L. (1992). An overview of Taguchi method and newly developed statistical methods for robust design. IIE

Transactions, 24 (5), 44–57. http://dx.doi.org/10.1080/07408179208964244

Inzinerine Ekonomika-Engineering Economics, 2014, 25(3), 320–332

- 332 -

Corrado lo Storto

Dviejų etapų metodo taikymas automobilių rinkos lyginamajai analizei

Santrauka

Technikos pramonės gaminių palyginimas gali būti vertinga priemone, padedančia automobilių pramonės gamintojams sustiprinti savo inovacinę

veiklą, numatyti perspektyvą kaip pagerinti savo gamybą ir padaryti ją konkurencinga ir patrauklia rinkoje. Gaminių palyginimas leidžia kompanijoms ne

tik palyginti ar įvertinti savo gaminių savybes ir veiklą su konkuruojančių kompanijų gaminių savybėmis ir veikla, tačiau ir įvertinti įdiegtos technologijos šiuolaikiškumą, taip pat gerinti jų projektavimą, gamybos procesą, rinkodaros strategiją ir taip įgyti tvirtą konkurencinį pranašumą.

Šiame darbe pasiūlytas dviejų etapų metodas: a) leidžiantis palyginti gaminius automobilių pramonėje, prieš tai įvertinus jų techninę vertę, kurią

automobilis suteikia vartotojams ir b) išsiaiškinti, kaip tai siejasi su automobilio pirkimo kaina ir naudojimo kaštais. Šioje analizėje, automobilis yra suvokiamas kaip techninių charakteristikų rinkinys, kuris suteikia tam tikras funkcijas ir kurios yra vertinamos

vartotojų. Techninės charakteristikos asocijuojasi su pačiu gaminiu, o funkcionalumo savybės asocijuojasi su gaminio veiksmu. Techninės

charakteristikos apima visas technologines posistemes ir sudėtines dalis, kurios yra automobilyje, t. y. variklio tipas, pakabos tipas, transmisijos sistema, oro kondicionavimo įranga ir t. t. Visos posistemės ir sudėtinės dalys veikia pagal tam tikrus techninius principus ir projektavimo taisykles. Šios

techninės charakteristikos yra inžinierių pasirinkimo, technologinės pažangos, automobilio kūrimo laiko ir automobilių pramonėje pritaikytų geriausių

inžinerinių praktikų rezultatas. Funkcionalumo savybės atlieka daug funkcijų, kurias automobilis suteikia vartotojams, t. y. mobilumas, saugumas, komfortas, kokybė ir t. t. Dažniausiai viena techninė charakteristika daro įtaką kartais vienai, kartais daugiau nei vienai funkcinei savybei. Iš tiesų,

vartotojai renkasi ir perka greičiau jau gaminio charakteristikas, negu patį gaminį, kuris tiesiog yra laikomas tarsi „juoda dėže“. Dažnai vartotojai visiškai

ignoruoja automobilio technologinius komponentus ir sistemas bei tai, kaip jie veikia visi kartu. Keleivinių automobilių vartotojams ne taip svarbu ar automobilyje įmontuotas keturių ar šešių cilindrų variklis, tačiau variklio galia arba greičio matai tikrai yra labai svarbūs jiems renkantis konkretų

automobilio modelį ir nusprendus jį pirkti. Metodas panaudoja seriją funkcinės veiklos (FV), siejamos su automobilio funkcinėmis savybėmis (FS), kurią

įvertinti yra daug lengviau negu technines charakteristikas. Funkcinės veiklos punktai yra sugrupuojami į keletą funkcinės veiklos kategorijų (FVK), kurios įvertina homogeniškų funkcijų, kurias automobilis suteikia vartotojams, veiklą. Šių savybių įvertinimai parodo automobilio naudotojams siūlomos

naudos kiekybinę indikaciją ir tuo pat metu netiesiogiai, automobilio modelyje apjungtų technologijų pradinę kiekybinę indikaciją. Pirmame etape

metodas pritaiko Duomenų apsupties analizę (DAA), kad apskaičiuotų santykinį techninį automobilio modelio efektyvumą (ATE), nes svertinis naudos ir kaštų santykis yra siejamas su serija funkcinių savybių, kategorijos veiklos įvertinimų, galinčių apibūdinti konkretų automobilio modelį (variklis,

mobilumas, saugumas, ir kokybė), o kaštai yra įvertinami pagal automobilio pirkimo ir naudojimo kaštus. Norint sukurti palyginimui rodiklį, yra

naudojamas automobilių modelių pavyzdys. Antrame etape, norint išsiaiškinti ryšį tarp automobilių modelių techninės vertės įvertinimo ir susijusių ekonominių kintamųjų egzistavimo, atliekama nelinijinė analizė, tam panaudojant ATE įvertinimą Techninė vertė įvertinama kaip automobilio teikiamos

naudos funkcija. Nors pirmame etape gaminių palyginimas turi mikroanalitinę perspektyvą, tačiau šiuo atveju sutelkiamas dėmesys į atskirus automobilių

modelius ir atsižvelgiama į jų charakteristikų specifinius įvertinimus. Antrame etape palyginimas turi makroanalitinę perspektyvą, nukreiptą į bendrų rinkos krypčių nagrinėjimą.

Metodas buvo įdiegtas norint ištirti rodiklius Italijos vietinių keleivinių automobilių rinkoje 1970-1990 metais. Pirmame etape rezultatai parodė, kad

keleiviniai automobiliai, parduoti Italijos rinkoje nuo aštuntojo iki dešimtojo dešimtmečio, labai skyrėsi vertinant jų techninį efektyvumą. Tik 35 automobilių modeliai paimti pavyzdžiais, DAA buvo klasifikuoti kaip 100% efektyvūs. Šioje efektyvių automobilių grupėje yra modelių, kurie priklauso

keliems rinkos segmentams, t. y. miesto arba sportiniai automobiliai. Iš tikrųjų, 100% efektyvūs automobilių modeliai yra brangesni už neefektyvius

automobilius. Didelis automobilių modelių skaičius (90) pasiekia aukštą techninio efektyvumo laipsnį, nuo 100% iki 82.91%, kur paskutinis efektyvumo įvertinimas yra „vidutinis efektyvumo matmuo“ kitame, mažesniame pavyzdyje, kuris apima tik 181, o ne 100% efektyvių automobilių modelių.

Keleivinių automobilių modeliai, parduoti rinkoje devintajame dešimtmetyje, buvo ne tokie konkurencingi techninio efektyvumo prasme kaip modeliai,

parduoti aštuntajame ir dešimtajame dešimtmečiuose. Iš tiesų, tik 4% keleivinių automobilių yra 100% efektyvūs. Nuo aštuntojo dešimtmečio pabaigos iki devintojo dešimtmečio vidurio Europos ir JAV automobilių gamintojai investavo daug pinigų, kad padidintų gaminių efektyvumą ir kokybę savo

gamyklose ir pateiktų rinkai geresnių gaminių. Metodas taip pat pateikė naudingas įžvalgas apie konkurencijos esmę Italijos automobilių rinkoje nuo

aštuntojo iki dešimtojo dešimtmečio. Buvo nustatytos septynios rinkos nišos, įskaitant automobilių modelius, turinčius tam tikras funkcinių savybių kombinacijas, kurios daro juos unikaliais Italijos automobilių rinkoje.

Antrame etape buvo atliktas lyginamasis tyrimas bendresniu lygiu, turint tikslą nustatyti bendras kryptis, kurios galėtų būti naudingos rinkodaros

profesionalams, inžinieriams ir projektuotojams, kuriant geresnį ir sėkmingesnį gaminį automobilių pramonėje. Rezultatai parodė, kad automobilio techninė vertė didėja, kai didėja automobilio pirkimo kaina, bet mažėja, kai kyla automobilio naudojimo kaštai.

Pasiūlytas metodas buvo pritaikytas norint atlikti retrospektyvinę Italijos keleivinių automobilių rinkos analizę, tačiau ir objektyvių ir subjektyvių

rodiklių panaudojimas, norint geriau įvertinti efektyvumą, kurį automobilis suteikia vartotojui, ir kaštų parametrus, o ne technines charakteristikas, apjungtas automobilio modelyje, daro šį metodą tinkamu taikyti, net jei būtų analizuojamos dabartinės pasaulio automobilių rinkos, turinčios savos šalies

specifiką. Siūlomas metodas kaip papildomą vertę siūlo naudingas įžvalgas: a) kaip palyginti automobilius, vertinant kelių matmenų savybes erdvėje,

vartotojams prieš tai pateikiant objektyvias techninės vertes; b) kaip analizuoti technologines kryptis automobilių pramonėje; c) kaip nagrinėti automobilių pramonės struktūrą ir nustatyti rinkos nišas, kurių dar neeksploatuoja automobilių gamintojai, atsiradimą. Tokio metodo taikymas gali būti

labai naudingas rinkodaros ir techniniams vadovams priimant teisingus sprendimus ir planus. Metodo privalumą sudaro didelis taikymo paprastumas, nes jis panaudoja nemažai objektyviai ir subjektyviai vertinamų funkcinių savybių bei kaštų parametrų, kurių dauguma yra pateikiami techninėje literatūroje.

Dar daugiau, metodas yra labai lankstus ir gali lengvai įtraukti tolesnius funkcinius instrumentus. Tai toks rodiklis, kuriuo įvertinama ar automobilis yra

tausojantis aplinką (tai nebuvo įtraukta į analizuojamą Italijos automobilių rinkos atvejį, nes tai nebuvo aktualu pavyzdyje įtrauktiems to meto automobilių modeliams, parduotiems rinkoje). Taip pat, šį metodą galima pritaikyti ir kitose srityse, pvz.: lėktuvų, kompiuterių ir spausdintuvų, namų

ūkio prekių, mobiliųjų telefonų ir t.t.

Raktažodžiai: duomenų apsupties analizė, automobilis, palyginimas, efektyvumas, gaminio vertė, pirkimo kaina, naudojimo kaštai.

The article has been reviewed.

Received in October, 2012; accepted in June, 2014.